The market put in an incredible reversal today with a huge intraday swing which was even larger than yesterday’s -6.66% loss. Not surprisingly, the extreme down and back up volatility has not happened very often in the past. However, there have been similar instances, and they have all led to a re-test of the bottom.
Unfortunately, I tried exactly modeling the past two day’s market action, and there are only several comparable instances. I had to loosen the criteria a bit, and even then, I have only come up with 5 other occurrences. Describing recent market action as unprecedented would not be inappropriate.
Anyway, here is what I used to model the Devil’s Bottom.
Buy Rules:
- Buy SPY if Yesterday Closed Down More than -2% and the Close was a 100 Day Low AND Today Closed Up More than 2% and Today’s Volume was More than Yesterday’s Volume
- Sell X Days Later
- No Commissions or Slippage Included
- All SPY History Used
A quick word about my methodology: Yesterday’s volume was incredible, but if I get too specific trying to get a near match of yesterday’s volume, it limits the sample size. Therefore, I just required today’s volume to be more than yesterday’s. It seems to work as anytime the market is putting in new 100 day lows, volume is swelling.
I also tried requiring a new 200 day low (yesterday was a new 200 day low) but that limited sample size. Again, I sufficed for the less restrictive 100 day low.
I also tried messing around with the relationships between yesterday’s and today’s opens and closes. Again, it was just too restrictive. As it stands, even with the less restrictive buy rules above, there were only 5 samples, not including today.
The Results:
Bottom Line:
Every instance saw a re-test of lows.
I’m not at all confident that these results are generalizable due to the small sample size.
What I do know is that the market is still in an abnormal state in a climate of high volatility with a 200 day moving average that is rolling over. An abnormal market will do abnormal things, and even if volatility starts to decline, it will decline with ever smaller spikes and waves.
My money is on a re-test. As I am writing this, I am working through some possibly more effective and generalizable ways of determining whether a re-test is likely. Stay tuned!
You have rally hit your stride with this sort of research. It is very interesting.
Thanks, Blogfather.
we are going back to soemwhere near the scene of the crime….
don’t we always go back to the scene of the crime? march 09 was deeply oversold, low vix relative to previous (much higher) lows, and a complete capit.
people still held hope after yesterday (monday)… we go back.
Agreud, but like everything else of the last 2 years, it is likely to fool people as it is happening.
Really like your style
Thanks!
Woody!
I still think I have a better shot at making money at my local casino bingo hall! I love it that you love stats like moi! Indeud!
You sure aug 9 volume should be greater than aug 8? From what I see, NYSE and $spx show greater volume on the aug 8. Also- the volume numbers are very similar (within 5-10% of each other) so perhaps it really doesn’t matter which day had greater volume.
I only look at SPY volume since that is what is actually tradeable, and it shows about 7 mil more shares traded than yesterday. But your point is valid. Maybe that requirement is too restrictive. Tomorrow I will try adjusting it so that it falls within a small percentage above or below. Thanks!
What about this:
Range today and range yesterday greater than 3x average daily range of last 100 days. Yesterday close down 2%. Volume yesterday and volume today greater than … something (maybe 2x 50d average?
Just thinking out load. Not tryin to step on anyone’s toes.
Love the site.
Thanks MIke. Like the range idea. I thought about trying that tonight but I was pretty sure it would be too restrictive. But as I think about it, it may be more appropriate.
Good stuff!
Shed, your skill at breaking down historical market data is always top notch and greatly appreciated!
Thanks guys. I’ve been working more tonight on new 200 day lows. Some interesting stuff. I will publish it tomorrow.
Great stuff, Wood – and thanks for the heads up to rein in my enthusiasm.
BTW, the GOP held what they needed to in Wisconsin – turnout was very heavy. Now it’s time to go back to building jobs instead of political ads.
Thanks Wood. You’re doing all of the work on ideas that cross my mind during the trading day. It’s really appreciated. Thanks again.
Woodshedder, just started reading your blog and it’s the finest quant research available gratis. I am a long time PPT member and can’t believe I just started read you.
This was an epic post that pointed out the danger, hopefully IBC readers were less hopeful in market direction: http://ibankcoin.com/woodshedderblog/2011/08/02/6-consecutive-lower-closes-whats-next/
Excuse the ignorance, but I am assuming number of days is trading days not calendar? In which case 11 days from that Tuesday post would be 8/17 as the worst average historical low.
There are strong political undercurrents in recent years with the market being used as leverage on politicians and the PPT trying to boost consumer confidence with the market. Tough to model, but the political angle has a strong effect. Would it be possible to model the days leading up to three day weekends when the market is depressed (currently looks like a great setup going into labor day weekend). With people anticipating the longer holiday they are typically naturally upbeat, and politicians would like the populous to gather with their family and not talk about how bad the economy / market is. Economists would like consumers to spend more on the weekends and a three day weekend has a greater opportunity for spending IF they feel confident?
Thanks AP! I want to answer your comment with some detail, but this market has kicked my butt. Must get sleep. Soon. I’ll be back in the a.m. with some thoughts.
Thanks for this site. Got it at Zero from you.
Great, glad to see you here on one of the finest trading blogs on the internet Paul. JakeGint’s blog is great for silver/gold as he follows those pretty closely – assuming if you follow ZH you might follow those too.
Bought twice at 1120 on S@P. Looks like major support since it hit it twice this week.